lwin_wkt_lex.c erroneously overridden

[aperi2007]

Hi, using the svn r8642, on a RH AS6 64bit , postgres 9.1.2 and geos 3.3.1 the make check phase report an error on the ticket test.

 tickets .. failed (diff expected obtained: /tmp/pgis_reg_10293/test_62_diff)

I add the logs reported on this ticket.

[aperi2007]

error log

[robe]

Andrea,

I'm not seeing an error. Can you try 3.3.2 SVN. I think some fixes were made in that. The error you are getting doesn't look related to my current cleanup changes.

[aperi2007]

Hi Robe,

I try recompiling all using the last geos-svn (3.4.0-svn) available.

But the error is still here.

I add the new log of error.

I use also gdal 1.9.0RC1, proj 4.7.0

[aperi2007]

The new error log with geos-svn

[aperi2007]

I forget to report the postgis-svn version: r8647.

[aperi2007]

Actully my settings for compile on RH 64bit are these:

CFLAGS="-m64" CPPFLAGS="-m64" CXXFLAGS="-m64" LDFLAGS="-m64" FFLAGS="-m64" LDFLAGS="-L/usr/lib64/" ./configure --with-pgconfig=/usr/local/pgsql/bin/pg_config --with-geosconfig=/usr/local/bin/geos-config --with-projdir=/usr/local/share/proj --with-xsldir=/usr/share/sgml/docbook/xsl-stylesheets-1.75.2 --with-libiconv=/usr/bin --with-xml2config=/usr/bin/xml2-config --with-gui --with-topology=yes --with-gdalconfig=/usr/local/bin/gdal-config --with-raster=yes --with-jsondir=/usr/local/lib

and gcc: 4.4.6 20110731 (Red Hat 4.4.6-3)

Now I'll try using a more simply version of settings.

[aperi2007]

Hi, I'm analyzing the failed test.

The failed test is this:

ST_Segmentize(ST_GeomFromEWKT('PolyhedralSurface( ((0 0 0, 0 0 1, 0 1 1, 0 1 0, 0 0 0)), ((0 0 0, 0 1 0, 1 1 0, 1 0 0, 0 0 0)), ((0 0 0, 1 0 0, 1 0 1, 0 0 1, 0 0 0)), ((1 1 0, 1 1 1, 1 0 1, 1 0 0, 1 1 0)), ((0 1 0, 0 1 1, 1 1 1, 1 1 0, 0 1 0)), ((0 0 1, 1 0 1, 1 1 1, 0 1 1, 0 0 1)) )'), 0.5);

I notice a difference with the others test in the same tickets set.

All the other call to ST_GeomFromEWKT will have the SRID explicitly declared.

ST_GeomFromEWKT('SRID=4326;POLYGONM(...

This instead no.

ST_GeomFromEWKT('PolyhedralSurface?(...

I guess is here the problem. Now try to do some tests....

[aperi2007]

Hi, I found the BUG !

Is the lovercase "PolyhedralSurface?" changing it to "POLYHEDRALSURFACE" all uppur the test pass correctly.

[robe]

Hmm that is strange. Wonder why its ot failing fo rme. It coud be a change I made. I know when I removed some of the deprecated functions I had to change a whole bunch of non-deprecated that were using deprecated to use the non-deprecated variant. It's a bit disturbing that it didn't trigger an error in any of my regress though. I'll look back and see what I changed. Might be I accidentally changed one to use ST_GeometryType by accident that was using GeometryType? (which though the name would suggest is deprecated is not).

[aperi2007]

I found some-else strange.

If I try the ST_GeomFromEWKT('PolyhedralSurface?...

On the windows experimental version it work without any problem.

If I try it on the redhat version it report the error.

I guess there a different on how parse the strings. On windows is case insensitive on redhat is case-sensitive ?

[robe]

hmm but that doesn't explain why Hudson wasn't registering errors either and I think its running red-hat though might be running a 32-bit. For the experimental I haven't updated that since my last sweep of cleanup so you might be testing against a newer Linux build. Was waiting till we call Alpha or by Friday whichever comes first before I release the next experimental build.

Windows shouldn't be anymore case insensitive for string compare than redhat unless there is a switch somewhere turned on causing it.

strk,dustymugs,pramsey -- are you noticing any regress failure with regular postgis regress?

[dustymugs]

I'm not seeing any regression failures using r8668 on 32-bit Slackware 13.37, 64-bit Slackware64 13.37 and OSX 10.6.8.

Slackware 13.37

CFLAGS="-g -O0" ./configure --with-raster

PostgreSQL 9.1.2 on i686-pc-linux-gnu, compiled by gcc (GCC) 4.5.2, 32-bit
Postgis 2.0.0SVN - 2012-01-04 15:25:33
  GEOS: 3.3.1-CAPI-1.7.1
  PROJ: Rel. 4.7.1, 23 September 2009

Slackware64 13.37

CFLAGS="-g -O0" ./configure --libdir=/usr/local/lib64 --with-raster

PostgreSQL 9.0.6 on x86_64-unknown-linux-gnu, compiled by GCC gcc (GCC) 4.5.2, 64-bit
Postgis 2.0.0SVN - 2012-01-04 15:33:55
  GEOS: 3.3.1-CAPI-1.7.1
  PROJ: Rel. 4.7.1, 23 September 2009

OSX 10.6.8

CFLAGS="-g -O0" ./configure --libdir=/usr/local/lib64 --with-gettext=/sw --with-raster

PostgreSQL 9.0.6 on x86_64-apple-darwin10.8.0, compiled by GCC i686-apple-darwin10-gcc-4.2.1 (GCC) 4.2.1 (Apple Inc. build 5666) (dot 3), 64-bit
Postgis 2.0.0SVN - 2012-01-04 15:41:51
  GEOS: 3.3.1-CAPI-1.7.1
  PROJ: Rel. 4.7.1, 23 September 2009

[strk]

Andrea: is it possible that your postgis source tree isn't clean ? I've seen the lexer/parser files being regenerated often on client machines. May affect casing.

[aperi2007]

Hi strk,

I guess to exclude this. My firstly test was to drop all the tree of postgis and reload it.

[robe]

Could be a language difference thing maybe that is causing case sensitivity.

Andrea, What version of libiconv are you using? I know you've flagged issues before that we couldn't see because your platform has non-English notices etc.

[aperi2007]

Replying to robe:

Could be a language difference thing maybe that is causing case sensitivity. Andrea, What version of libiconv are you using? I know you've flagged issues before that we couldn't see because your platform has non-English notices etc.

My RH will use iconv (GNU libc) 2.12

[strk]

Andrea: please double check with 'svn status'. See #1439 where the reporter confirmed it was a non-clean tree.

Note that I'm not saying you intentionally changed the sources, but I think I've seen some files being regenerated when not needing to...

[aperi2007]

Hi strk, you have reason !

here the results.

$ svn status
?       doc/topology_comments.sql
?       doc/postgis_cheatsheet.html
?       doc/raster_cheatsheet.html
?       doc/topology_cheatsheet.html
?       doc/tiger_geocoder_cheatsheet.html
?       doc/tiger_geocoder_comments.sql
?       doc/raster_comments.sql
M       liblwgeom/lwin_wkt_lex.c
$

I see the liblwgeom/lwin_wkt_lex.c was modified (??)

After see this I drop all the sources and svn flags. And reload from zero all. But this Modificed files there is again.

I don't know why.

[strk]

Let's find out !

[strk]

I tried hard but couldn't reproduce the accidental override. Andrea: I'll close this but please file a new one if you can figure out _what_ overrides that file.